Visualization

rm(list = ls())
library(tidyverse)
library(broom)
library(MASS)
library(usmap)
library(RColorBrewer)

Data transformation and filtering

• Following Louise’s methods: this section is exactly the same as hers except added ID to variable list
• I recoded loan status = “Bad” ~ 1, “Good” ~ 0
• Didn’t dichotomize tot_coll_amt
• Anlysis using complete cases (i.e., NAs excluded)

load("/Users/carlyb/Desktop/Fall2018/BST260/loan.RData")

#recode variables
dat <- loan.dat %>%
   mutate(perc_funded_amnt_inv = funded_amnt_inv/funded_amnt,
         issue_d = as.character(issue_d),
         term = as.character(term),
         year = as.numeric(str_sub(issue_d, start = -4)))

varlist <- c('loan_status', 'loan_amnt', 'funded_amnt',
                'int_rate', 'grade',
                'emp_length', 'home_ownership',
                'annual_inc', 'verification_status',
                'purpose',
                'addr_state', 'dti',
                'delinq_2yrs', 'inq_last_6mths',
                'open_acc', 'pub_rec', 
                'revol_bal', 'revol_util',
                'total_acc', 'initial_list_status',
                'application_type', 'acc_now_delinq',
                'tot_coll_amt', 'tot_cur_bal',
                'total_rev_hi_lim', 'perc_funded_amnt_inv',
                'term', 'year','id')

dat <- dat[, varlist]

#filter out issued, regroup loan_status
dat <- dat %>%
  filter(!loan_status == "Issued") %>%
  mutate(loan_status_bin = case_when(loan_status %in% c('Fully Paid', 'Current', 'Does not meet the credit policy. Status:Fully Paid') ~ "Good", loan_status %in% c("Default","Charged off","Does not meet the credit policy. Status:Fully Paid","Grade Period","Late (31-120 days)","Late (16-30 days)") ~ "Bad"))

#replacing NAs with the mean value 
#for (j in 1:ncol(dat)){
#  miss = is.na(dat[,j])
#  if (sum(miss) > 0){
#    dat[miss, j] = mean(dat[,j], na.rm=T)}}

#recode string variables to numbers: employment length and grade 
letters <- LETTERS[1:26]
dat <- dat %>%
  mutate(emp_length_2 = gsub("years|year|<|\\+","", emp_length) %>% as.numeric(),
         grade_2 = match(grade, letters),
         status_dum = ifelse(loan_status_bin == "Good",0,1),
         term_dum = ifelse(term == "36 months",0,1))

## Warning in function_list[[k]](value): NAs introduced by coercion

#funModeling::df_status(dat, print_results = FALSE)

US Map Plots

• Proportion of loan = ‘bad’ by state
• Loan grade by state

#table(dat$grade)
color_blind_friendly_cols <- c("#999999", "#E69F00", "#56B4E9", 
                               "#009E73", "#F0E442", "#0072B2", 
                               "#D55E00", "#CC79A7")

#this plot will show the proportion of loan defaults by state
state.dat  <- dat %>%
  group_by(addr_state) %>%
  summarize(state.prop.default =  mean(status_dum, na.rm = TRUE),
            state.prop.gradeA = sum(grade == "A")/n(),
            state.prop.gradeB = sum(grade == "B")/n(),
            state.prop.gradeC = sum(grade == "C")/n(),
            state.prop.gradeD = sum(grade == "D")/n(),
            state.prop.gradeE = sum(grade == "E")/n(),
            state.prop.gradeF = sum(grade == "F")/n(),
            state.prop.gradeG = sum(grade == "G")/n()) %>%
  mutate(state = addr_state)

#head(state.dat)

plot_usmap(data = state.dat, regions = "state", values = "state.prop.default") +
  scale_fill_continuous(name = "") + 
  ggtitle("Proportion of Loan Defaults by State") + 
  theme(legend.position = "right")

#this plot will show the most common loan grade for each state
grade.state.dat <- state.dat %>%
  group_by(addr_state) %>%
  mutate(max.val = max(state.prop.gradeA,state.prop.gradeB,   state.prop.gradeC,state.prop.gradeD,state.prop.gradeE,state.prop.gradeF,state.prop.gradeG),
         maxGrade = ifelse(max.val == state.prop.gradeA, "A", NA),
         maxGrade = ifelse(max.val == state.prop.gradeB, "B", maxGrade),
         maxGrade = ifelse(max.val == state.prop.gradeC, "C", maxGrade),
         maxGrade = ifelse(max.val == state.prop.gradeD, "D", maxGrade),
         maxGrade = ifelse(max.val == state.prop.gradeE, "E", maxGrade),
         maxGrade = ifelse(max.val == state.prop.gradeF, "F", maxGrade),
         maxGrade = ifelse(max.val == state.prop.gradeG, "G", maxGrade))

plot_usmap(data = grade.state.dat, regions = "state",values = "maxGrade") +
  scale_fill_brewer(name = "",palette = "Blues") +
  ggtitle("Most Common Loan Grade by State") +
  theme(legend.position = "right")

Loan defaults over time

dat %>%
  group_by(year) %>%
  summarize(prop.default =  mean(status_dum, na.rm = TRUE)) %>%
  ggplot(aes(year,prop.default)) +
  geom_point() +
  geom_line() +
  scale_y_continuous(name = "Proportion Defaulted", breaks = seq(0,0.1,0.02),limits = c(0,0.1)) +
  scale_x_continuous(name = "Year") +
  ggtitle("Proportion Defaulted over Time")

Loan defaults over time by grade

dat %>%
  group_by(year,grade) %>%
  summarize(prop.default =  mean(status_dum, na.rm = TRUE)) %>%
  ggplot(aes(year,prop.default,color = grade)) +
  geom_point() +
  geom_smooth(se = F) +
  scale_y_continuous(name = "Proportion Defaulted", breaks = seq(0,0.1,0.02),limits = c(0,0.1)) +
  scale_x_continuous(name = "Year") +
  ggtitle("Proportion Defaulted over Time") +
  scale_color_brewer(palette = "Blues", name = "LendingClub Loan Grade") +
  theme_dark()

## `geom_smooth()` using method = 'loess' and formula 'y ~ x'

## Warning: Removed 161 rows containing missing values (geom_smooth).

Loan defaults over time by state

#probably don't want to use this one. sqrt transform is not great, and log transform gives tons of NAs from 0s.
dat %>%
  group_by(addr_state,year) %>%
  summarize(state.prop.default =  mean(status_dum, na.rm = TRUE)) %>%
  ggplot(aes(year,addr_state,fill = state.prop.default)) +
  geom_tile(color = "grey50") +
  scale_x_continuous(expand=c(0,0)) +
  scale_fill_gradientn(colors = brewer.pal(9, "Blues"), trans = "sqrt", name='Square Root of Proportion') +
  theme_minimal() +  theme(panel.grid = element_blank()) +
  ggtitle("Proportion of LendingClub Loan Defaults over Time by State") + 
  ylab("") + 
  xlab("")

Loan Amount vs. DTI

#take random sample to be more readable
#still working on this one
set.seed(12218)
#sumdat <- dat %>%
    #group_by(grade) %>%
   # summarize(grp_dti = mean(dti, na.rm = T),
             # grp_loan_amnt = mean(loan_amnt, na.rm = T)) %>%
    #left_join(dat, by = "grade")

dat %>%
    sample_n(10000) %>%
    ggplot(aes(dti,loan_amnt,color = grade)) +
    geom_point(alpha = 0.25) + 
    geom_smooth(se = F) +
    ylab("Loan Amount") +
    xlab("Debt-to-Income Ratio") +
    ggtitle("DTI vs. Loan Amount by LendingClub Loan Grade") +
    scale_color_discrete(name = "Grade")

## `geom_smooth()` using method = 'gam' and formula 'y ~ s(x, bs = "cs")'

Amount funded vs. Loan Grade

dat %>%
  ggplot() + 
  geom_boxplot(aes(y = loan_amnt, x = grade, color = grade)) +
  scale_color_brewer(name = "LendingClub Grade",palette = "Blues") +
  theme_dark() +
  scale_y_continuous(name = "Loan amount") +
  scale_x_discrete(name = "Grade") +
  ggtitle("Distributions of Loan Amount ($) by LendingClub Loan Grade")

Amount funded vs. income

xbrk <- rep(5,50)
xbrk <- xbrk^(order(xbrk))
ybrk <- rep(2,50)
ybrk <- ybrk^(order(ybrk))
set.seed(12218)
dat %>%
    sample_n(10000) %>%
    ggplot(aes(annual_inc,loan_amnt)) +
    geom_point(alpha = 0.25) + 
    scale_y_continuous(name = "Log Loan Amount", trans = "log",breaks = ybrk) +
    scale_x_continuous(name = "Log Annual Income (USD)", trans = "log", breaks = xbrk) +
    ggtitle("Income vs. Loan Amount")

Distribution of Debt to Income Ratio

dat %>%
  filter(!is.na(loan_status_bin)) %>%
  ggplot() + 
  geom_boxplot(aes(y = dti, x = loan_status_bin,color = grade)) +
  scale_color_brewer(name = "LendingClub Grade",palette = "Blues") +
  theme_dark() +
  scale_y_continuous(name = "Debt-to-Income Ratio") +
  scale_x_discrete(name = "Grade") +
  ggtitle("Distributions of DTI by LendingClub Loan Grade and Loan Status")

#excluding 2 outliers
dat %>%
  filter(dti < 7500 & !is.na(loan_status_bin)) %>%
  ggplot() + 
  geom_boxplot(aes(y = dti, x = loan_status_bin,color = grade)) +
  scale_color_brewer(name = "LendingClub Grade",palette = "Blues") +
  theme_dark() +
  scale_y_continuous(name = "Debt-to-Income Ratio") +
  scale_x_discrete(name = "Grade") +
  ggtitle("Distributions of DTI by LendingClub Loan Grade and Loan Status")

#sqrt transform
dat %>%
  filter(!is.na(loan_status_bin)) %>%
  ggplot() + 
  geom_boxplot(aes(y = sqrt(dti), x = loan_status_bin,color = grade)) +
  scale_color_brewer(name = "LendingClub Grade",palette = "Blues") +
  theme_dark() +
  scale_y_continuous(name = "Debt-to-Income Ratio") +
  scale_x_discrete(name = "Grade") +
  ggtitle("Distributions of DTI by LendingClub Loan Grade and Loan Status")

#overall distribution
dat %>%
  filter(!is.na(loan_status_bin)) %>%
  ggplot() + 
  geom_boxplot(aes(y = sqrt(dti), x = loan_status_bin,color = loan_status_bin)) +
  scale_color_brewer(name = "Loan Status",palette = "Blues") +
  theme_dark() +
  scale_y_continuous(name = "Debt-to-Income Ratio") +
  scale_x_discrete(name = "Loan Status") +
  ggtitle("Distributions of DTI by Loan Status")

Distribution of Loan Amount grouped by outcome

dat %>%
  filter(!is.na(loan_status_bin)) %>%
  ggplot() + 
  geom_boxplot(aes(y = sqrt(loan_amnt), x = loan_status_bin,color = loan_status_bin)) +
  scale_color_brewer(name = "Loan Status",palette = "Blues") +
  theme_dark() +
  scale_y_continuous(name = "Loan Amount (USD)") +
  scale_x_discrete(name = "Loan Status") +
  ggtitle("Distributions of Loan Amount by Loan Status")